key: cord-0865100-lnls2mhg
authors: Boettler, Tobias; Csernalabics, Benedikt; Salié, Henrike; Luxenburger, Hendrik; Wischer, Lara; Alizei, Elahe Salimi; Zoldan, Katharina; Krimmel, Laurenz; Bronsert, Peter; Schwabenland, Marius; Prinz, Marco; Mogler, Carolin; Neumann-Haefelin, Christoph; Thimme, Robert; Hofmann, Maike; Bengsch, Bertram
title: SARS-CoV-2 vaccination can elicit a CD8 T-cell dominant hepatitis
date: 2022-04-21
journal: J Hepatol
DOI: 10.1016/j.jhep.2022.03.040
sha: 7bb612d68c9aa14932782f523d738da7f7d1b500
doc_id: 865100
cord_uid: lnls2mhg

BACKGROUND & AIMS: Autoimmune hepatitis episodes have been described following SARS-CoV-2 infection and vaccination but their pathophysiology remains unclear. Here, we report the case of a 52-year-old male, presenting with bimodal episodes of acute hepatitis, each occurring 2-3 weeks after BNT162b2 mRNA vaccination and sought to identify the underlying immune correlates. The patient received first oral budesonide, relapsed, but achieved remission under systemic steroids. METHODS: Imaging mass cytometry for spatial immune profiling was performed on liver biopsy tissue. Flow cytometry was performed to dissect CD8 T cell phenotypes and identify SARS-CoV-2-specific and EBV-specific T cells longitudinally. Vaccine-induced antibodies were determined by ELISA. Data was correlated with clinical labs. RESULTS: Analysis of the hepatic tissue revealed an immune infiltrate quantitatively dominated by activated cytotoxic CD8 T cells with panlobular distribution. An enrichment of CD4 T cells, B cells, plasma cells and myeloid cells was also observed compared to controls. The intrahepatic infiltrate showed enrichment for CD8 T cells with SARS-CoV-2-specificity compared to the peripheral blood. Notably, hepatitis severity correlated longitudinally with an activated cytotoxic phenotype of peripheral SARS-CoV-2-specific, but not EBV-specific CD8+ T cells or vaccine-induced immunoglobulins. CONCLUSIONS: COVID19 vaccination can elicit a distinct T cell-dominant immune-mediated hepatitis with a unique pathomechanism associated with vaccination induced antigen-specific tissue-resident immunity requiring systemic immunosuppression. LAY SUMMARY: Liver inflammation is observed during SARS-CoV-2 infection but can also occur in some individuals after vaccination and shares some typical features with autoimmune liver disease. In this report, we show that highly activated T cells accumulate and are evenly distributed in the different areas of the liver in a patient with liver inflammation following SARS-CoV-2 vaccination. Moreover, within these liver infiltrating T cells, we observed an enrichment of T cells that are reactive to SARS-CoV-2, suggesting that these vaccine-induced cells can contribute to the liver inflammation in this context.

Background & Aims: Autoimmune hepatitis episodes have been described following SARS-CoV-2 infection and vaccination but their pathophysiology remains unclear. Here, we report the case of a 52-year-old male, presenting with bimodal episodes of acute hepatitis, each occurring 2-3 weeks after BNT162b2 mRNA vaccination and sought to identify the underlying immune correlates. The patient received first oral budesonide, relapsed, but achieved remission under systemic steroids.

Methods: Imaging mass cytometry for spatial immune profiling was performed on liver biopsy tissue. Flow cytometry was performed to dissect CD8 T cell phenotypes and identify SARS-CoV-2-specific and EBV-specific T cells longitudinally. Vaccine-induced antibodies were determined by ELISA. Data was correlated with clinical labs.

Results:Analysis of the hepatic tissue revealed an immune infiltrate quantitatively dominated by activated cytotoxic CD8 T cells with panlobular distribution. An enrichment of CD4 T cells, B cells, plasma cells and myeloid cells was also observed compared to controls. The intrahepatic infiltrate showed enrichment for CD8 T cells with SARS-CoV-2-specificity compared to the peripheral blood. Notably, hepatitis severity correlated longitudinally with an activated cytotoxic phenotype of peripheral SARS-CoV-2-specific, but not EBV-specific CD8+ T cells or vaccine-induced immunoglobulins.

Conclusions: COVID19 vaccination can elicit a distinct T cell-dominant immune-mediated hepatitis with a unique pathomechanism associated with vaccination induced antigen-specific tissue-resident immunity requiring systemic immunosuppression.

Liver inflammation is observed during SARS-CoV-2 infection but can also occur in some individuals after vaccination and shares some typical features with autoimmune liver disease.

In this report, we show that highly activated T cells accumulate and are evenly distributed in

Vaccination is the key strategy to fight the global COVID19 pandemic. There was no hepatitis safety signal in COVID19 vaccination trials [1] , however several reports have recently associated autoimmune hepatitis (AIH)-like conditions with COVID19 vaccines [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] . To our knowledge, no severe case with liver failure requiring liver transplantation was reported. Liver injury was observed after both, mRNA and vector-based vaccines, while time from vaccine administration to symptom onset ranged between 4 days after the first dose to 6 weeks after the second dose. One patient was re-exposed to the vaccine which led to a worsening of liver injury [11] . It remains unclear whether the reported association of autoimmune hepatitis with vaccination is coincidental, might reflect transient drug-induced liver injury, or could involve unique SARS-CoV-2-induced antigen-specific immune activation [12] . However, the fact that AIH-like conditions also occurred after SARS-CoV-2 infection [13] suggests that the latter could be a driving factor for the sporadic cases.

Here, we describe the case of a 52-year-old male presenting with acute mixed hepatocellular/cholestatic hepatitis after the first dose of BNT162b2 mRNA vaccine and severe hepatitis after the second dose. Diagnostic evaluation was compatible with criteria for autoimmune hepatitis (AIH). After initiation of oral budesonide therapy, liver function tests improved for a month before a relapse occurred that was successfully treated with systemic prednisolone and ursodeoxycholic acid. Comprehensive immunologic assessment of the inflammatory infiltrates in the liver revealed the presence of a highly activated cytotoxic CD8 T cell infiltrate including a SARS-CoV-2-specific CD8 T cell population that correlated with the peripheral activation of SARS-CoV-2-specific CD8 T cells, indicating that post-COVID19 vaccination hepatitis involves vaccination-elicited antigen-specific immune responses with distinct histological features compared to bona fide autoimmune hepatitis.

The 52-year-old male patient with no remarkable medical history other than pre-existing hypothyroidism under long-term substitution therapy with levothyroxine and normal historic liver function tests (LFT) developed progressive nausea, fatigue, loss of appetite and pruritus with symptoms starting approximately 10 days after the first (prime) dose of the BNT162b2 mRNA vaccine. He subsequently developed jaundice and presented at his primary care physician with LFT indicative of acute mixed hepatocellular/cholestatic hepatitis (ALT: 2130 U/l, AP: 142 U/l gamma-GT: 217 U/l, Bilirubin 7.7 mg/dl) (Fig. 1) . The patient was admitted to a primary care center 25 days post first vaccination. Viral hepatitis A, B, C and E as well as cytomegalovirus-and Epstein-Barr virus-infections were excluded by serology and/or PCR J o u r n a l P r e -p r o o f testing. HFE genotyping did not reveal hemochromatosis associated variations. There was also no significant alcohol consumption and autoimmune serology remained inconclusive with borderline AMA-M2 reactivity. The patient recovered rapidly without specific therapy and was discharged with decreasing LFTs after three days under leading differential diagnosis of a toxic hepatitis. Over the next two weeks, liver enzymes declined further, with normalization of AST and AP and the patient received his second (boost) dose of the BNT162b2 mRNA vaccine 41 days after the first vaccination (Fig. 1 ). 20 days post boost vaccination (dpb), the patient reexperienced nausea and fatigue. Lab testing revealed a relapse of acute mixed hepatitis with (ALT 1939 U/l, ALP 167 U/l, bilirubin 2.9 mg/dl). He was subsequently referred to our tertiary center at 26dpb. Autoimmune serology was repeated with mild Hyperglobulinemia (IgG levels 1.02-fold of ULN, normal IgA and IgM levels), ANA (1:200) and borderline positivity for antismooth muscle antibodies and AMA-M2 while tests for anti-LKM remained negative. We performed a liver biopsy that histologically showed interface hepatitis with a moderate degree of lymphoplasmacytic infiltrate and foci of lobular necrosis and apoptosis. Eosinophilic granulocytes were not present. Neither relevant perisinusoidal nor portal fibrosis were observed. Together, these findings are compatible with a probable autoimmune hepatitis according to the revised original score for autoimmune hepatitis [14] and the patient was treated with 9 mg oral budesonide per day. Over the next weeks liver enzymes declined before a relapse occurred 39 days after initiation of therapy (66 dpb), that was controlled subsequently after therapy escalation to systemic steroids in combination with ursodeoxycholic acid. The patient's LFTs subsequently normalized within 8 weeks (Fig. 1 ). Anti-spike antibodies did not show major fluctuations with similar titers compared to healthy individuals at the time of diagnosis at 27 days post boost vaccination (Fig. 1B) and an expected reduction of titers over time ( Fig 1C) .

To understand the immune infiltrate in the liver in more detail, in addition to conventional histochemistry, we performed highly multiplexed imaging mass cytometry with a broad panel These findings suggested a different immune cell contribution compared to typical spontaneous AIH. We next assessed the spatial localization of the different immune cell subsets within the liver parenchyma. The strongest immune infiltration was observed in periportal areas (Fig. 2C) . Notably, while B cells and plasma cells were also enriched but predominantly present in periportal areas, CD8 T cell distribution was panlobular (Fig. 2C, D) .

To test for immune correlates of hepatic damage, we assessed the cytotoxic granule marker Granzyme B. Interestingly, we observed a strong accumulation of cytotoxic (Granzyme Bpositive) CD8 T cells while other Granzyme B-expressing immune cell subsets, such as granulocytes, were not increased (Fig. 2E) . These data are also in line with our observation of increased intrahepatic expression of CD38, HLA-DR and transcription factor T-bet as markers of activated effector T cells ( Fig. 2A , F). Based on their strong enrichment, widely scattered distribution and activated cytotoxic phenotype within the liver parenchyma, we speculated that CD8 T cells could be drivers of the hepatic inflammation.

We next performed a detailed analysis of the intrahepatic and peripheral CD8 T cell populations using flow cytometry. The patient's intrahepatic CD8 T cell population was enriched for markers of tissue residency (CXCR6, CD69 and CD103) and activation (CD38) (Fig. 3A) . Expression of CD38 on CD8 T cells was also observed in the peripheral blood.

Interestingly, CD38 expression levels were markedly elevated in the patient compared to postvaccination time-point matched control vaccinees that did not develop hepatitis (75.9% vs 15.4%, respectively) (Fig. 3B) . The unusual expansion of activated cytotoxic T cells observed in our spatial analysis led us to hypothesize that vaccination-induced SARS-CoV-2 spikespecific CD8 T cells might contribute to liver disease as we have recently observed rapid induction of SARS-CoV-2 spike-specific CD8 T cells by the BNT162b2 mRNA vaccine [15] .

Indeed, using a patient-matched HLA-A*03 tetramer loaded with a SARS-CoV-2-spike epitope (S378-386) we identified spike-specific CD8 T cells. In the peripheral blood, spike-specific CD8 T cells were 10.2fold more abundant than T cells specific for an EBV-specific CD8+ T cell control epitope (Suppl. Fig. 1 ). We also identified spike-specific CD8 T cells in the liver. Compared to the peripheral blood, the intrahepatic CD8 T cell pool was ~3.4 fold enriched for spike-specific CD8 T cells (Fig. 3C ) and displayed tissue-resident characteristics (CXCR6, CD103, CD69) and strong activation as indicated by CD38 expression (Fig. 3C ). Spike protein was not identified in the liver at the time of analysis by immunohistochemistry (data not shown). We next asked whether frequency and activated phenotype of circulating spike-specific CD8 T cells was maintained during therapy. Interestingly, longitudinal analyses showed stable J o u r n a l P r e -p r o o f frequencies of circulating spike-specific CD8 T cells (Fig. 3D ) with decreasing CD38 expression levels after the initiation of budesonide therapy and coinciding with declining transaminase levels. However, CD38 expression on spike-specific CD8 T cells and other cytotoxicity-associated markers such as GzmB and T-bet increased when the patient relapsed under budesonide therapy, then normalized after introduction of systemic immunosuppressive therapy (Fig. 3E-G) . This pattern was not observed for LMP-1419-427-loaded HLA-A*24:02restricted EBV-specific CD8 T cells (Supp Fig 1) . It was also not observed for bulk CD8 T cell analyses, that however showed a high abundance of activated cytototoxic CD8 T cell populations in the periphery (Fig. 3E-G) . In sum, these data indicate broad activation of cytotoxic CD8 T cells with the peripheral phenotype of spike-specific CD8+ T cells reflecting the clinical response pattern to immunosuppressive therapy. coincidence. In particular, AIH-like disease after vaccination was reported in patients with age and gender characteristics typical for spontaneous AIH [6, 7, 10] . While some of these cases thus may represent coincidence, a causal relationship to the vaccine is also possible, such as bystander hepatitis driven by elevation of systemic cytokines or chemokines after vaccination, similar to cases occurring in association with natural SARS-CoV-2 infection [13] . The varying patterns of clinical manifestation and the wide range of time elapsed between vaccine administration and symptom onset clearly suggest that different mechanisms may contribute to these reported cases. Here, our analysis highlights that activated cytotoxic CD8 T cells including vaccine-induced spike-specific CD8 T cells could contribute to disease pathogenesis.

Importantly, autoimmune hepatitis is a condition that requires lifelong immunosuppressive therapy in many affected patients [16] . It is therefore important to differentiate AIH from possibly transient immune-mediated hepatitis post vaccination. Diagnosis of autoimmune hepatitis is typically established using AIH scoring tools informed by liver function tests, auto-antibody serology and typical histological features such as interface hepatitis and enrichment of plasma cells [14] . In our case, prior to therapy, AIH diagnosis was deemed "probable" based on the modified original AIH score. It is important to note in this context that the first episode of acute hepatitis after the first vaccine dose was self-limiting without therapy, which also guided our choice to initiate therapy with budesonide in an attempt to minimize systemic side effects and to maintain systemic anti-SARS-Cov2 immunity if possible. However, our patient experienced a relapse 3-4 weeks after budesonide therapy and then required systemic steroid therapy, that was empirically combined with ursodeoxycholic acid, a combination used in the therapy of autoimmune hepatitis overlap syndrome that was chosen due to presence of AMA-M2 antibodies typically found in primary biliary cholangitis [17] . The patient subsequently recovered quickly, with only a gingivitis occurring during systemic steroid therapy as a possible immunosuppression-related event. Due to relapse hepatitis after steroid tapering, the patient was put under long-term maintenance immunosuppressive therapy under which he achieved complete biochemical remission.

Mechanistically, the disease manifestation in our patient was distinct from classical spontaneous AIH, that is typically associated with elevated peripheral immunoglobulins, a plasma cell dominated infiltrate and prominent interface hepatitis. Here, while there was a slight elevation of peripheral immunoglobulins and intrahepatic enrichment of intrahepatic B cells and plasma cells, a more striking correlate was observed at the level of cytotoxic CD8 T cells.

Activated cytotoxic CD8 T cells were strongly enriched in the patient liver to an extent that they represented the most abundant intrahepatic immune cell population and included vaccine elicited SARS-CoV2 specific responses. Notably, the peripheral activation state of these spikespecific CD8 T cells correlated with hepatitis severity and the clinical course after introduction of immunosuppressive therapy. These results implicate T cells as a key pathogenic immune cell type of this vaccine-associated immune hepatitis as a novel subtype of autoimmune hepatitis. The early increase in ALT values after first and second dose of BNT162b2 and the observation that CD8 T cells including spike-specific CD8+ T cells dominate the immune infiltrate also fits to recent observations that demonstrate an early mobilization of spike-specific CD8 T cells already after the first dose of the vaccine [15] . Of note, our tetramer reagent likely stains only a fraction of vaccination-induced antigen-specific T cells, fitting to the broad expression of T cell activation and cytotoxicity markers observed in the analysis of the bulk CD8 T cell population. However, it is also possible that the broad activation pattern observed involves other, non-SARS-CoV2 specific "bystander" CD8 T cell populations. Moreover, the precise mechanism that causes infiltration of the spike-specific CD8 T cells into the liver which adopt features of tissue resident memory T cells (TRM) remains unclear. Virus-specific CD8+ T cells can accumulate in the liver even if the primary site of infection is distant, such as during influenza infection and it is possible that the liver can act as a "graveyard" for activated T cells [18, 19] . However, activated cytotoxic CD8 T cells can mediate hepatitis even in the absence of antigen [20] . In addition, TCR-independent, innate-like cytotoxicity of bystander T cells has been described in acute infections [21, 22] . Of note, CXCR6+ CD8+ TRM cells, such as those found in this patient, may mediate auto-aggression in metabolic liver diseases, in response to local inflammatory cues [23, 24] . Whilst these reports raise the possibility of an antigenindependent T cell-mediated hepatitis, another possible explanation could be the presence of their cognate antigen, i.e. the spike protein. While we could not detect spike protein by IHC in the liver, it has to be noted that biopsy was performed 27 days after the second vaccine dose and transient expression after vaccination, which could have caused CXCR6-expressing spike-specific CD8 T cells to home to the liver and target antigen-presenting cells, cannot be excluded. However, additional mechanisms, such as antigen cross-recognition may also contribute to the immunopathology. In sum, BNTb163b2 vaccine may trigger immune- 

Liver tissue was obtained by transcutaneous biopsy, formalin-fixed, paraffin-embedded, cut into 4 µm sections and stained as previously described. [25] Briefly, after deparaffinization, rehydration, antigen-retrieval and blocking, slides were stained with metal-labeled antibodies and air-dried. Image acquisition of the biopsy (30.912 mm 2 ) and 2.25 mm 2 of the control samples was performed using a Hyperion Imaging System (Fluidigm; USA). ROIs were laserablated spot-by-spot at 200 Hz, resulting in a pixel resolution of 1 µm 2 . Image visualization was performed with FIJI (v1.52p, ImageJ). Segmentation of single cells was conducted using an adapted unbiased, supervised analysis pipeline using Ilastik (v 1.3.3) and CellProfiler (v 3.1.9).

To analyze liver zones the distance of each pixel to GLUL, CD34 and α-SMA was calculated using CellProfiler (v 4.0.4), added to the high-dimensional information of each cell and used for gating in OMIQ (Omiq Inc). Liver zones were defined: Centrilobular (distance to GLUL ≤50µm), periportal (distance to GLUL >50 µm and distance to CD34 and α-SMA ≤50 µm) and intermediate (distance to GLUL >50 µm and distance to CD34 and α-SMA >50 µm). Absolute cell counts were normalized to 1 mm². Stacked barplots were created with R version 4.0.1 using ggplot2.

Peripheral blood mononuclear cells (PBMCs) were isolated from anticoagulated blood samples after density gradient centrifugation and subsequently stored at -80°C until further usage.

Remaining liver tissue was mechanically homogenized and filtered through a 70 µm cell strainer (Corning). Cells were washed and directly processed for flow cytometry analysis.

Additionally, freshly isolated PBMCs were also stained and analyzed by flow cytometry. 

The generated flow cytometry and imaging mass cytometry datasets will be made available in accordance with legal regulations upon reasonable request. Arrowheads indicate timepoint of liver biopsy. Duration and dose of budesonide (9 mg/d) and

prednisolone therapy (initially 50 mg/d) is shown in orange and red colorization. B: Comparison of anti-spike S1 IgG antibody titers in the patient to timepoint-matched vaccines who did not develop hepatitis. C: Longitudinal analysis of anti-spike S1 IgG antibodies before and after the initiation of therapy. 

Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine

Letter to the editor: Autoimmune hepatitis after COVID-19 vaccination. A rare adverse effect? Hepatology

Autoimmune hepatitis developing after coronavirus disease 2019 (COVID-19) vaccine: Causality or casualty

Autoimmune hepatitis developing after coronavirus disease 2019 (COVID-19) vaccine: Causality or casualty?

An Anusual Case of Acute Cholestatic Hepatitis After m-RNA BNT162b2 (Comirnaty) SARS-CoV-2 Vaccine: Coincidence, Autoimmunity or Drug Related Liver Injury

Another case of autoimmune hepatitis after SARS-CoV-2 vaccination. Still casualty?

The mRNA COVID-19 vaccine -a rare trigger of Autoimmune Hepatitis

Autoimmune hepatitis following SARS-CoV-2 vaccine: May not be a casuality

Liver injury following SARS-CoV-2 vaccination: a multicenter case series

Autoimmune hepatitis following COVID-19 Vaccination: true causality or mere association

Immune-mediated hepatitis with the Moderna vaccine, no longer a coincidence but confirmed

Autoimmune Hepatitis Developing After Coronavirus Disease 2019 (COVID-19) Vaccine: One or Even Several Swallows Do Not Make a Summer

Autoimmune hepatitis triggered by COVID-19: A report of two cases

International Autoimmune Hepatitis Group Report: review of criteria for diagnosis of autoimmune hepatitis

Rapid and stable mobilization of CD8+ T cells by SARS-CoV-2 mRNA vaccine

EASL Clinical Practice Guidelines: Autoimmune hepatitis

Diagnosis and Management of Primary Biliary Cholangitis

Selective retention of activated CD8+ T cells by the normal liver

Characteristics of virus-specific CD8(+) T cells in the liver during the control and resolution phases of influenza pneumonia

Cytokine-dependent bystander hepatitis due to intrahepatic murine CD8 T-cell activation by bone marrow-derived cells

Innate-like Cytotoxic Function of Bystander-Activated CD8(+) T Cells Is Associated with Liver Injury in Acute Hepatitis A

Contribution of herpesvirus specific CD8 T cells to anti-viral T cell response in humans

Auto-aggressive CXCR6(+) CD8 T cells cause liver immune pathology in NASH

NASH limits antitumour surveillance in immunotherapy-treated HCC

Deep spatial profiling of human COVID-19 brains reveals neuroinflammation with distinct microanatomical microglia-T-cell interactions

J o u r n a l P r e -p r o o f